Cognitive Mechanisms underlying Visual Perspective Taking in typical and ASC children

Transcription

1 Cognitive Mechanisms underlying Visual Perspective Taking in typical and ASC children Amy Pearson, 1. Lauren Marsh, 2. Danielle Ropar, 3. and Antonia Hamilton, Psychology Department, University of Durham. Mountjoy Science Site, Durham, DH1 3LE. 2. School of Experimental Psychology, University of Bristol, Bristol, BS8 1TU. 3. School of Psychology, University of Nottingham, University Park, Nottingham, NG7 2RD. 4.Institute of Cognitive Neuroscience, UCL, 17 Queens Square, London, WC1N 3AR. Running Title: Cognitive Mechanisms of Perspective Taking Manuscript info: 5272 words 4 figures 2 tables Correspondence Details: Amy Pearson Psychology Department University of Durham Durham DH1 3LE Phone: (+44) amy.pearson@durham.ac.uk

2 1 Lay Abstract Research has suggested that people with Autism Spectrum Conditions (ASC) may find it difficult to see things from different points of view (visual perspective taking). This experiment aimed to investigate why this is and whether children use different strategies in perspective taking tasks. Sixty children with and without ASC took part. Each child completed a perspective taking task in which they had to decide what a toy on a table would look like from different points of view; a mental rotation task in which they decided how a toy would look after it had been turned round; and a body matching task in which they had to match pictures of a person shown from different angles. Results showed that children with ASC performed better than typically developing children at mental rotation and at the same level in visual perspective taking and body matching. In children without autism, the ability to take another perspective was linked to how good they were at the body matching task, whereas in the ASC children perspective taking was linked to how good they were at mentally rotating objects. These results suggest that children with and without autism use different strategies when confronted with a visual perspective taking task, and are able to achieve similar levels of performance in different ways.

3 2 Scientific Abstract Previous research has suggested that people with Autism Spectrum Conditions (ASC) may have difficulty with visual perspective taking (VPT), but it is not clear how this relates to different strategies that can be used in perspective taking tasks. The current study examined VPT in 30 children with autism and 30 verbal mental age matched typical children, in comparison to mental rotation abilities and body representation abilities. Using a similar paradigm to Hamilton et al. (2009) all children completed three tasks: a VPT task in which children decided what a toy on a table would look like from a different points of view; a mental rotation task in which the child decided what a toy would look like after it had been rotated; and a body posture matching task, in which children matched pictures of a body shown from different viewpoints. Results showed that children with ASC performed better than the TD children on the mental rotation task, and at a similar level on the VPT task and body matching task. Importantly, in the typical children VPT performance was predicted by performance on the body matching task, whereas in the ASC children VPT performance was predicted by mental rotation ability. These findings suggest that differences in VPT in ASC may be explained by the use of a spatial rotation strategy rather than the embodied egocentric transformation strategy used by typical children. Keywords: Autism Spectrum Conditions, Visual Perspective Taking, Mental Rotation, Embodied, Cognitive Mechanisms.

4 3 Introduction When two travellers peer at a map from different locations, both can see the streets but it may take extra consideration to realise that ahead to one viewer is left to the other. Visual perspective taking (VPT) is the ability to consider another person s viewpoint on the world and is traditionally divided into level 1 VPT (can she see the object?) and level 2 VPT (what does it look like to her?) (Flavell, Everett, Croft, & Flavell, 1981). VPT2 is the process which the two map-readers must engage in to communicate effectively it draws on both spatial skills to consider the map and social skills to consider what representations are in the other s mind. Recent research has shown that children with autism spectrum conditions (ASC) perform worse than expected on a VPT2 task (Hamilton, Brindley, & Frith, 2009) compared to typically developing (TD) children. In the present paper, we explore and expand on this result with a new study which examines the strategies underlying VPT performance in typical and autistic children. Taking another visual perspective is a complex task which draws on a number of different cognitive processes (Surtees, Apperly, & Samson, 2013a). Different people (or the same person in a different context) may use different cognitive strategies to accomplish the same task (Gardner, Brazier, Edmonds, & Gronholm, 2013; Kessler & Wang, 2012). In the VPT2 task used by Hamilton et al. (2009) children aged 4-12 years were shown four pictures of a toy (e.g. a cow) from four canonical orientations and asked to point to the picture that matched the orientation of the same toy on the table. The real toy was then covered, and a doll was placed to the left or right of the toy. The child was then asked which cow will the doll see? and answered by pointing to one of the cow pictures. To give a successful response on this task, the child could adopt a strategy of an embodied egocentric

5 4 transformation (EET), and imagine herself in the place of the doll in order to see the world through the doll s eyes. This strategy draws on the ability to manipulate body representations and may be related to other social skills (Kessler & Rutherford, 2010; Kessler & Wang, 2012). Alternatively, the child could adopt a strategy of mental rotation (MR), and imagine the hidden toy turning around so that the side that was in front of the doll is now in front of the child. She can now consider her own new view of the imagined toy to answer the question (Zacks & Tversky, 2005). This strategy draws on the ability to mentally transform objects in space. Both strategies can lead to the correct answer in this task, but they draw on quite different cognitive systems (Surtees et al., 2013a). There are a number of reasons to believe that people with autism may find EET difficult. Autism is characterised by difficulties with social cognition, in particular theory of mind (ToM) (Frith, 2001, 2012; Happe, 1995; Senju, 2012). Neuroimaging (Schurz, Aichhorn, Martin, & Perner, 2013) and cognitive (Surtees, Apperly, & Samson, 2013b) studies suggest links between the ability to consider another person s thoughts and taking their visual perspective. Previous studies of VPT2 in ASC did not examine the specific strategy children might use. Out of three studies conducted on VPT2 in autism, two studies have reported poor VPT2 in children with ASC (Hamilton et al., 2009; Yirmiya, Sigman, & Zacks, 1994) while one reported intact performance (Tan & Harris, 1991). In their 2009 study, Hamilton et al. examined VPT2 alongside MR and ToM ability in children with autism and a group of verbal mental age (VMA) matched TD children. They found that in TD children, VPT2 performance is predicted more strongly by ToM ability than it is by MR skills and verbal IQ. Results also showed a task by group interaction, where children with ASC were significantly worse on the VPT2 trials than the typical children, but

6 5 performed better on the MR task. However, floor effects in this study make it difficult to make strong claims about the direction of results. In this paper our aim is to expand on Hamilton s past work and examine the strategies which typical and autistic children might use to perform VPT2. Two secondary aims are to replicate previous findings (Hamilton et al., 2009) without floor effects, and investigate whether manipulating the wording of the test question, in regards to VPT2 for self and other viewpoints would impact on performance. The current study is concerned with two possible strategies a MR strategy and an EET strategy. Several studies suggest that typical adults use an EET strategy to perform VPT2 tasks (Surtees et al., 2013a; Yu & Zacks, 2010). This process involves representing the body posture and position of the target and then mentally transforming the self to match that target (Grush, 2004; Kessler & Thomson, 2009). Body information is critical in this process (Kessler & Thomson, 2009). Thus, if children use an EET strategy to perform VPT2, we would expect their performance to correlate with their ability to perform other types of body transformation. To measure body transformation abilities in children, we use a posture matching task (Pearson, 2014). In this task, children must match a photo of a person in a particular posture to a photo of the same person in the same posture taken from a different viewpoint. To solve the task, the child must create a viewpoint independent representation of the body posture and manipulate it. We predict that children who are good at body posture matching will also be good atvpt2, if those children use an EET strategy. This is likely to be the case for the TD children based on previous research (Zacks & Tversky, 2005). However, previous research has indicated that people with autism may have impaired body representations (Eigsti, 2013) and are impaired at EET (Pearson, Marsh, Hamilton, & Ropar, 2014). Thus, children with ASC may find it hard to use an EET strategy. An

7 6 alternative strategy that children could use is to perform MR on the scene (Zacks & Tversky, 2005). Zacks and Tversky (2005) found that typical adults could use a MR strategy to complete a perspective taking task, but this strategy was less efficient than performing an EET. A child using a MR strategy could ignore body postures and simply imagine the scene rotating until the part nearest the other viewer is closest to the child. This strategy is very similar to the control task of MR used here and previously (Hamilton et al., 2009). We predict that children who are good at MR will also be good at VPT2 if those children use a MR strategy. This is likely to be the case for the children with ASC based on the assumption that they find EET problematic (Pearson et al., 2014). To summarise, we predict a relationship between VPT2 performance and body representation in the TD group, and a relationship between VPT2 performance and MR in the ASC group. In addition to examining the strategy used to perform VPT2, we were interested in whether manipulating the test question would impact on performance in the ASC and TD children. VPT2 studies typically ask about what another person would see from a different viewpoint, but participants could also be asked what would you see if you were at a different viewpoint. Considering the mental states of another and the mental states of the self may draw on similar cognitive processes (Frith & Happe, 1999). Here we test if this applies to VPT. Previous studies in TD adults have shown little difference behaviourally between the ability to see things from someone else s point of view versus seeing things for oneself from a new point of view (Kessler & Thomson, 2009) as they both require the simultaneous representation of two different viewpoints. However, these different subtypes of VPT2 (VPT2 for self and other) have not been examined in people with ASC. It is possible that those with ASC might find it easier to represent their own view from another location than

8 7 imagining another person s viewpoint. Alternatively, they might find it equally difficult as judging another person s visual perspective. Thus, we modified Hamilton s VPT2 task to include two different conditions. One measured perspective taking for another person, VPT2 other (VPT2O). This was used in the original study ( what will Suzy see? ). Additionally, we added a condition to measure perspective taking for self, VPT2 self (VPT2S), asking what would you see if you were sitting over there. This meant that it was possible to examine whether these subtypes of VPT2 were different in children with and without autism. Based on previous findings we predict that children with ASC will be impaired on VPT2 tasks compared to the TD children (as both require an EET), but that MR performance will be intact.

9 8 Method Participants Sixty children participated in this study. Thirty children with a diagnosis of ASC were recruited from schools in Nottinghamshire and Wales. Their mean chronological age (CA) was 9.27 years and 27 were male. The British Picture Vocabulary Scale (BPVS) (Dunn, Dunn, Whetton, & Burley, 1997) was used to establish VMA, and the Social Communication Questionnaire (SCQ) (Berument, Rutter, Lord, Pickles, & Bailey, 1999) and Social Aptitude Scale (SAS) (Liddle, Batty, & Goodman, 2009) were completed by a caregiver to evaluate the child s social understanding and communication skills. All of the ASC children had a previous diagnosis from an independent clinician, confirmed by the parent/caregiver in a background questionnaire. The task was also completed by 30 VMA matched TD children (see Table 1). They had a mean CA of 6.83 years and 18 were male. The TD children were recruited during Nottingham University s Summer Scientist Week, an event where children take part in several research studies. All TD children completed the BPVS and their caregiver completed the SAS. None of the typical children had a diagnosis of ASC or any other learning difficulty, confirmed by parent questionnaire. All parents of participating children and their schools consented to taking part in the study, which was approved by The University of Nottingham ethics committee. ----TABLE 1----

10 9 Design & procedure A repeated measures design was used to examine the effects of task on performance (here measured in terms of accuracy). Each child completed four experimental tasks: MR, VPT2 self (VPT2S), VPT2 other (VPT2O) and body representation. Performance on each task was measured by calculating number of trials correct, which was transformed into a percentage. Children with ASC also completed a ToM battery and their parents completed the SCQ/SAS. The ASC children were tested individually in a quiet room at school or at home, whereas the TD children were tested individually in a quiet, partitioned cubicle at the Summer Scientist event. The tasks administered were: VPT2 and MR tasks: These tasks were closely based on Hamilton et al, Materials were a small turntable, an opaque pot and three toys (a bear, a frog and a small fire truck). The turntable was marked with a square with different colours on each side (See Figure 1). The experimenter sat beside the child at the table, and three empty chairs marked with coloured stickers were placed around the table. At the start of each trial, the toy was placed on the turntable facing one of the coloured strips. The child held a picture card showing four images of the toy from different viewpoints and was asked which picture can you see? (Figure 1a). This established that the child was attending to the initial orientation of the toy. For the VPT2S trials the toy was covered with an opaque pot and the child asked if you were sitting at the [blue] side of the table (indicating the empty chair with a blue sticker), which picture would you see when I lift up the pot? (Figure 1b). For the VPT2O task the toy was covered with the opaque pot and a doll was placed at another side of the table. The child was asked, Jim is sitting on the [blue] side of the table, when I lift the pot up which picture will Jim see? (Figure 1c). Other

11 10 colours were substituted as appropriate, to test the alternative viewpoints. For the MR trials the toy was then covered with an opaque pot, and rotated to a different orientation. The child was then asked when I lift the pot up, which picture will you see? (Figure 1d). In all trials, the child could respond by selecting the corresponding picture on the answer card (Figure 1e). Praise was given for all answers Figure Each child completed six trials of the VPT2O task, six trials of the VPT2S task and six trials of the MR task. Trials were blocked by task, and task order was counterbalanced across participants. For the VPT2 tasks the six trials presented were a selection of the three different table viewpoints in a pseudo randomised order (each viewpoint was presented twice) used in combination with each of the four viewpoints of the toy (i.e. front of the toy is facing Jim, Jim is sat on the red side of the table). For the MR task the six trials presented were a pseudorandom selection of the four different viewpoints of the toy and four different starting points for rotation. For each correct answer a score of 1 was given and these were averaged to give a percentage of correct scores for each participant. Body representation task The body representation task assessed children s ability to match pictures of human body postures across different orientations. Both meaningful and meaningless postures were used to determine if meaning or familiarity impacts on performance, as previous studies have used a mixture of both meaningful and meaningless stimuli, leading to inconsistency in findings (Dowell, Mahone, & Mostofsky, 2009; Ham, Corley, Rajendran, Carletta, & Swanson, 2008; Hamilton, Brindley, & Frith, 2007; Reed et al., 2007). Stimuli defined as meaningful

12 11 depicted familiar postures, i.e. communicative postures such as extending an arm to communicate stop (see Figure 2a) whereas for meaningless stimuli, unfamiliar postures were used (i.e. a random limb configuration, see Figure 2b). Body pictures were generated by taking two simultaneous photographs of a clothed person in a distinctive body posture from two different locations. For each matched pair, a foil picture showing the same person performing a different posture was also presented. Pilot testing on adult participants was used to equate difficulty between the different stimuli (Pearson, 2014). Thus, stimuli on each trial consist of a trio of images exemplar, target and foil (Figure 2). These images were printed in colour on laminated cards. For each trial there were two cards, one depicting two body postures (one target match and one foil) and one depicting an exemplar to be matched (Figure 2). ----Figure On each trial, the child was first given a laminated card with two pictures (the target and foil) then given a second laminated card with a single picture (the exemplar). The experimenter asked which one of these (point to double picture card) matches your picture? The child could respond either verbally or non-verbally by pointing or putting the single card with the appropriate match. One practice trial with a different posture was given prior to the experimental trials, and any errors the child made were corrected with an explanation. After the child understood the task, the experimenter presented the 12 experimental trials (6 meaningful bodies, and 6 meaningless). Stimuli were presented in blocks because mixing meaningful and meaningless stimuli reduces the impact of meaning (Tessari & Rumiati, 2004). The order of trials within a block was pseudo-randomised across children and the order of blocks (meaningful and meaningless) was counterbalanced. Praise was given

13 12 throughout regardless of response. For each correct answer a score of 1 was given and these were averaged to give a percentage of correct scores for each participant. Theory of mind battery All ASC children were tested on their ToM ability. They were assessed on their understanding of diverse desires and beliefs, knowledge access, false belief, contents false belief and a penny hiding task (Baron-Cohen, Leslie, & Frith, 1985; Devries, 1970; Wellman & Liu, 2004; Wimmer & Perner, 1983). For each task, each child was given a score of 1 if they passed and 0 if they failed, with a maximum score of 12. This score was converted into a percentage correct for analysis. TD children were not tested for their ToM ability due to time constraints.

14 Results VPT and MR Performance The original study from Hamilton et al. (2009) compared the performance of an ASC and TD group on MR and VPT2O tasks. To examine whether results from this study replicated previous results, an Analysis of Covariance (ANCOVA) was used to examine performance on MR and VPT2O in the ASC and typical groups. Each child s score on the MR and VPT2O tasks were entered as repeated measures factor, with group, BPVS-raw score and SAS score as additional predictors. SAS was included in the analysis to test for relationships between parent-rated social function and our tasks. Results showed a marginal effect of group (F (1, 54) = 3.366, p=0.066) with the TD children performing worse than the ASC children (Figure 3) and a significant interaction between task and group (F (1, 57) =5.924, p=0.018). Here the typical children scored worse on MR compared to the ASC group (t (58) =-2.11, p=0.039) but showed similar performance on the VPT2 other task (t (58) =-.349, p=0.728). This replicates the results found in Hamilton et al, There was no effect of task and no interaction between task and BPVS. There was a marginal interaction between task and SAS (F (1, 54) =3.042, P=0.087) showing that accuracy increased with SAS score. There was a significant effect of BPVS, with those with higher BPVS scores performing better on the tasks (F (1, 54) =36.879, p<0.001). No further interactions were found Figure The current study included separate tasks to measure VPT2S and O. In order to examine whether performance on VPT2S and VPT2O was similar in the two groups, an ANCOVA was conducted with group as a between-subjects variable and BPVS raw score and SAS as

15 14 covariates. The ANCOVA showed that there was no significant effect of task (p=0.496) and no interaction between task and group (p=0.684), suggesting that VPT2 self and VPT2 other are very similar processes in both ASC and TD participants. To further investigate this relationship a bivariate correlation was performed, with VPT2O and VPT2S as inputs. This showed that VPT2S and O were highly correlated across children (r=.65, p<0.001), therefore they were collapsed to give a single VPT2 score for each child. This was used in further analysis. To determine the effect of group on VPT2 (overall) and MR performance, an ANCOVA with a between-subjects variable of group, task as a repeated measure and covariates of BPVS raw score and SAS was conducted. Results showed a significant effect of group (F (1, 54) =4.551, p=0.037) with the ASC group performing better than the TD group (Figure 3). There was a significant interaction between task and group (F (1, 54) =6.576, p=0.013) with the typical group showing poorer performance on MR than the ASC group (t (58) =-2.11, p=0.032), but no difference between groups on the VPT2 task (t (58) =-.431, p=0.668). There was a significant effect of task (F (1, 54) =5.330, p=0.025) with both groups more accurate on the VPT2 task than MR. There was also a significant effect of BPVS raw score (F (1, 57) =40.998, p<0.001) in that children with a higher BPVS raw score were more accurate but no interaction between task and BPVS (F (1, 54) =2.592, p=0.113). There was no significant effect of SAS (p=.204), however, there was a marginal interaction between task and SAS (F (1, 54) =3.214, p=0.079) showing that as SAS score increased, accuracy also increased across tasks.

16 15 Body Representation Task An ANCOVA was used to examine the effects of group and stimulus category (meaningful/meaningless) on accuracy, with raw BPVS and SAS entered as covariates. There was a significant effect of meaning (F (1, 54) = 8.31, p=0.006) with both groups showing higher accuracy for the meaningless stimuli (Figure 4) and a marginal effect of SAS (F (1, 54) =3.45, p=0.069) with higher SAS participants performing better than low SAS participants. There was a significant effect of BPVS (F (1, 54) = 18.84, p<0.001) with higher BPVS participants performing better. There were no significant effects of group and no interactions between any of the variables (all p>0.01) -----Figure Which Factors Predict VPT2 performance in ASC and TD children? Separate regression analyses were used to test which measures predicted VPT2 performance in the typical and ASC children. Data for the 30 TD children were entered into a multiple linear regression model testing how VPT2 was predicted by MR, body representation, SAS, BPVS raw score and age. The regression model had an overall fit of R²=.65. Performance on VPT2 was significantly predicted by performance on the BPVS (β=.385, p<0.038) and body representation task (β=.458, p=0.011) in the TD children. Data for the 30 ASC children were also entered into a multiple linear regression model to determine how VPT2 was predicted by MR, body representation, SAS, BPVS and age. The regression model had an overall fit of R²=.73, and VPT2 was significantly predicted by

17 16 performance on the BPVS (β=.473, p=0.012) and MR task (β=.661, p<0.001). A further regression analysis examined the additional variables collected only in the ASC group. Here ToM and SCQ were entered alongside MR, body representation, SAS, BPVS and age as predictors. The regression model had an overall fit of R²=.78, and VPT2 was significantly predicted by performance on the BPVS (β=.392, p=0.043), MR task (β=.597, p<0.001) and SCQ (β=-.311, p=0.048). Details of the regression analyses are presented in Table Table

18 17 Discussion The main aim of the current study was to investigate the cognitive processes involved in taking another person s visual perspective. Results showed that in children with autism, MR ability predicts VPT2 performance, whereas in typical children body representation ability predicts VPT2 performance. We also replicated the findings of Hamilton et al. (2009) without the floor effects. Here we consider each of our three tasks (VPT2, MR and body matching) individually and then consider what our results mean for overall theories of VPT and social cognition in autism. Individual tasks The VPT tasks required the child to consider what a toy looks like from another person s point of view (VPT2O) or what a toy would look like if the child were in a different place (VPT2S). Performance on the self and other tasks was highly correlated across children, suggesting that both types of VPT draw on the same cognitive processes in each child. This parallels findings for ToM, where imagining the mental states of others or the future self are similar (Frith & Happe, 1999). However, this does not mean that all children use the same strategy (see discussion of group differences in strategy below). Overall, children with and without ASC performed at a comparable level on the two VPT2 tasks. This is consistent with Hamilton et al, 2009, where performance was similar for children with ASC and VMA matched typical children. However, the present study avoids the floor effects seen in the previous study, and thus confirms more clearly that children with ASC can perform a VPT2 task at a level appropriate for their VMA.

19 18 The MR task requires the child to consider what a toy will look like after it has been rotated. Results from this task showed that the TD children performed significantly worse than the ASC children, which is consistent with previous work (Hamilton et al, 2009), but again avoids floor effects. This is also consistent with a recent meta-analysis (Muth, Honekopp, & Falter, 2014) and with previous studies showing that people with autism often display better performance on non-verbal measures of performance compared to their verbal ability (Joseph, Tager-Flusberg, & Lord, 2002). One alternative explanation for these differences between groups is the difference in gender ratio. There were more females in the TD group than the ASC group and previous research has shown that males tend to outperform females on MR (Geiser, Lehmann, & Eid, 2008). However, we found no difference in performance between male and female participants both within and across groups in the MR task. This makes it unlikely that gender was a stronger predictor of performance than group. Overall, we suggest that the ability to perform MR in ASC is better relative to younger mental aged matched controls. The body representation task required children to match images of body postures across different viewpoints. Results revealed no significant effect of group on performance, however there was a significant effect of SAS. Children with higher SAS scores (the majority were the TD children, see Table 1) were better at the body representation task, suggesting a relationship between social ability and difficulties representing the human body in 3D or matching bodies from different points of view. These findings are consistent with previous research showing a similar relationship between body representation and social ability in TD adults (Kessler & Wang, 2012) and indicate that social ability in general, beyond an autism

20 19 diagnosis, may be an important factor in predicting the ability to represent the body from different points of view. The results of this task also revealed that all children performed significantly better on the meaningless than the meaningful stimuli. These findings contrast with studies showing an advantage for processing meaningful stimuli in TD adults (Bosbach, Knoblich, Reed, Cole, & Prinz, 2006) in which prior knowledge of postures aids recognition. This difference may best be understood in terms of different effects of meaning in children and adults. The stimulus trios were piloted on adult participants and selected so that meaningful and meaningless trios were equally hard for adults. If adults show an advantage for meaningful stimuli (Bosbach et al., 2006), this selection procedure would give us meaningful trios which are intrinsically harder to match because adults can use their knowledge of the stimulus meaning to overcome the complexity. However, if children are not able to benefit from meaning in the same way as adults, they will find the meaningful stimuli harder, as our results show. In summary, the ASC children performed similarly to younger VMA matched typical children on both the VPT2 and body representation tasks. This performance suggests that these abilities are in line with their VMA (which was the same as the TD children). On the MR task the ASC children performed better than the TD children, suggesting that MR ability is better than those of a similar VMA. However, it would be inappropriate to suggest that MR performance is superior, due to this group having an overall higher CA. The inclusion of an age-matched control group on the VPT and MR tasks in future research would aid in clarifying the extent to which these skills are delayed or superior in ASC.

21 20 Predictors of VPT2 performance The design of the current study allows us to test how performance on a variety of tasks relates to VPT abilities. Our regression analyses examine how age, BPVS, MR, body representation, ToM, SAS and SCQ scores relate to VPT2. As BPVS was a consistent predictor across all groups, soaking up effects of age, we do not consider this further. Rather, we discuss how each of the other measures relates to VPT2, beyond the general effect of verbal IQ. We found that performance on the body posture task predicts VPT2 performance in the TD children but not in the ASC children. This suggests that typical children use a bodyrelated strategy to perform the VPT2 task. The EET strategy previously describe in adults is a strong candidate here (Kessler & Thomson, 2009). In this approach, the child imagines themselves in the bodily position and orientation of the doll in the VPT2 task, thus drawing on the same body representation skills as the posture matching task. This is consistent with previous research in adults (Kessler & Thomson, 2009; Surtees et al., 2013b) which suggests that in order to complete VPT2, TD people represent the body posture and position of the person with the target perspective and then mentally transform their own body to match this viewpoint. Examining the mental rotation task, we found that scores predicted VPT2 performance in the ASC children but not the TD children. This suggests that children with ASC use a MR strategy to perform VPT2, in which they mentally turn the toy from the doll s point of view to their own in order to complete the task. This means that the children with ASC are not using the (typical) EET strategy to perform the VPT2 task. Recent research has

22 21 shown that people with autism may have difficulty with using the self as a reference frame when performing spatial transformations (Pearson, Marsh, Hamilton and Ropar, 2014) and may draw upon spatial information in perspective taking tasks if it is available to them (Langdon & Coltheart, 2001). Overall, the present data are consistent with the claim that there are two possible strategies that can be used to accomplish a VPT2 task an EET strategy or a MR strategy. Typical children prefer to use the former, while ASC children prefer to use the latter. This implies that in tasks which can be solved using both a social and spatial strategy, people with ASC might be able to compensate for difficulty in social cognition if they have good spatial skills. However, the spatial strategy may be suboptimal in typical adults, MR strategies tend to be slower and less accurate than performing an EET (Zacks and Tversky, 2005). We can also consider how performance on the VPT2 task relates to ToM and everyday social skills (measured with the SCQ and SAS). In the previous study (Hamilton et al. 2009) there was a strong relationship between ToM ability and VPT2 performance in the TD children. This is consistent with earlier findings (Aichhorn, Perner, Kronbichler, Staffen, & Ladurner, 2006; Farrant, Fletcher, & Maybery, 2006; Flavell, 1988). In the current study we were only able to examine ToM in the ASC group and found no relationship between ToM and VPT2 ability. This is compatible with the claim that the ASC participants are using a different, spatial strategy to perform the VPT2 task which cannot help them perform the ToM tasks. Note that our study is correlational and does not show whether being more social makes a child better at VPT or having better VPT skills makes a child better at other social skills. However, it is possible that encouraging children with ASC to make use of body information and an EET strategy in VPT tasks could generalise to better use of VPT and ToM

23 22 in other contexts. A relationship was found between SCQ score and VPT2 ability in the ASC children; participants with better social skills also showed better VPT2 skills. This is consistent with previous studies (Dawson & Fernald, 1987). This could mean that good use of a spatial strategy helps children with autism in real-world social situations as measured on the SCQ, or could reflect individual differences in the use of social strategies among the ASC group tested. Conclusions This study tested children with ASC and VMA-matched typical children on VPT, body representation and MR tasks. Results indicate that typical children use an embodied egocentric transformation to perform VPT, drawing on their good body representation skills. In contrast, the children with autism may use a mental rotation strategy to perform the VPT task, drawing on their strong spatial skills. Our results emphasise the importance of considering different strategies in understanding spatial and social tasks, and may demonstrate compensatory processing in the children with autism.

24 23 Acknowledgements We would like to thank all of the schools, children and parents who took part in this research for giving us their time. Amy Pearson and Lauren Marsh were funded by the University of Nottingham.

25 References Aichhorn, M., Perner, J., Kronbichler, M., Staffen, W., & Ladurner, G. (2006). Do visual perspective tasks need theory of mind? NeuroImage, 30(3), Baron-Cohen, S., Leslie, A. M., & Frith, U. (1985). Does the autistic child have a "theory of mind"? Cognition, 21(1), doi: (85) [pii] Berument, S. K., Rutter, M., Lord, C., Pickles, A., & Bailey, A. (1999). Autism screening questionnaire: diagnostic validity. British Journal of Psychiatry, 175, Bosbach, S., Knoblich, G., Reed, C. L., Cole, J., & Prinz, W. (2006). Body inversion effect without body sense: Insights from deafferentation. Neuropsychologia, 44(14), doi: DOI /j.neuropsychologia Dawson, G., & Fernald, M. (1987). Perspective-Taking Ability and Its Relationship to the Social- Behavior of Autistic-Children. Journal of Autism and Developmental Disorders, 17(4), doi: Doi /Bf Devries, R. (1970). Development of Role-Taking as Reflected by Behavior of Bright, Average, and Retarded Children in a Social Guessing Game. Child Dev, 41(3), 759-&. doi: DOI /j tb01029.x Dowell, L. R., Mahone, E. M., & Mostofsky, S. H. (2009). Associations of Postural Knowledge and Basic Motor Skill With Dyspraxia in Autism: Implication for Abnormalities in Distributed Connectivity and Motor Learning. Neuropsychology, 23(5), doi: Doi /A Dunn, L. M., Dunn, L. M., Whetton, K., & Burley, J. (1997). British picture vocabulary scale (2nd ed). Windsor: NFER-NELSON. Eigsti, I. M. (2013). A review of embodiment in autism spectrum disorders. Front Psychol, 4, 224. doi: /fpsyg Farrant, B. M., Fletcher, J., & Maybery, M. T. (2006). Specific language impairment, theory of mind, and visual perspective taking: Evidence for simulation theory and the developmental role of language. Child Development, 77(6), doi: DOI /j x Flavell, J. H. (1988). The development of children s knowledge about the mind: From cognitive connections to mental representations: Cambridge University Press. Flavell, J. H., Everett, B. A., Croft, K., & Flavell, E. R. (1981). Young children's knowledge about visual perception: Further evidence for the Level 1â Level 2 distinction. Developmental Psychology, 17(1), doi: / Frith, U. (2001). Mind blindness and the brain in autism. Neuron, 32(6), doi: Doi /S (01) Frith, U. (2012). Why we need cognitive explanations of autism. Quarterly Journal of Experimental Psychology (Hove). doi: / Frith, U., & Happe, F. (1999). Theory of mind and self-consciousness: What is it like to be autistic? Mind & Language, 14(1), Gardner, M. R., Brazier, M., Edmonds, C. J., & Gronholm, P. C. (2013). Strategy modulates spatial perspective-taking: evidence for dissociable disembodied and embodied routes. Frontiers in Human Neuroscience, 7. doi: Artn 457 Doi /Fnhum Geiser, C., Lehmann, W., & Eid, M. (2008). A note on sex differences in mental rotation in different age groups. Intelligence, 36(6), Grush, R. (2004). The emulation theory of representation: motor control, imagery, and perception. Behavioural Brain Science, 27(3), ; discussion Ham, H. S., Corley, M., Rajendran, G., Carletta, J., & Swanson, S. (2008). Brief report: Imitation of meaningless gestures in individuals with Asperger syndrome and high-functioning autism.

26 25 Journal of Autism and Developmental Disorders, 38(3), doi: DOI /s x Hamilton, A. F., Brindley, R., & Frith, U. (2009). Visual perspective taking impairment in children with autistic spectrum disorder. Cognition, 113(1), doi: S (09) [pii] /j.cognition Hamilton, A. F. D., Brindley, R. M., & Frith, U. (2007). Imitation and action understanding in autistic spectrum disorders: How valid is the hypothesis of a deficit in the mirror neuron system? Neuropsychologia, 45(8), doi: DOI /j.neuropsychologia Happe, F. G. E. (1995). Understanding Minds and Metaphors - Insights from the Study of Figurative Language in Autism. Metaphor and Symbolic Activity, 10(4), Joseph, R. M., Tager-Flusberg, H., & Lord, C. (2002). Cognitive profiles and social-communicative functioning in children with autism spectrum disorder. Journal of Child Psychology and Psychiatry, 43(6), doi: / Kessler, K., & Rutherford, H. (2010). The Two Forms of Visuo-Spatial Perspective Taking are Differently Embodied and Subserve Different Spatial Prepositions. Front Psychol, 1, 213. doi: /fphys Kessler, K., & Thomson, L. A. (2009). The embodied nature of spatial perspective taking: Embodied transformation versus sensorimotor interference. Cognition, 114(1), Kessler, K., & Wang, H. F. (2012). Differently embodied transformations in visuo-spatial perspective taking. Cognitive Processing, 13, S21-S22. Langdon, R., & Coltheart, M. (2001). Visual perspective-taking and schizotypy: evidence for a simulation-based account of mentalizing in normal adults. Cognition, 82(1), Liddle, E., Batty, M., & Goodman, R. (2009). The Social Aptitudes Scale: an initial validation. Social Psychiatry and Psychiatric Epidemiology, 44(6), doi: /s Muth, A., Honekopp, J., & Falter, C. M. (2014). Visuo-spatial performance in autism: a meta-analysis. J Autism Dev Disord, 44(12), doi: /s Pearson, A. (2014). The Development of Visuo-Spatial Perspective Taking in Autism Spectrum Disorder. Unpublished Doctoral Dissertation, Nottingham, University of Nottingham. Pearson, A., Marsh, L., Hamilton, A., & Ropar, D. (2014). Spatial Transformations of Bodies and Objects in Adults with Autism Spectrum Disorder. J Autism Dev Disord. doi: /s Reed, C. L., Beall, P. M., Stone, V. E., Kopelioff, L., Pulham, D. J., & Hepburn, S. L. (2007). Brief report: perception of body posture--what individuals with autism spectrum disorder might be missing. Journal of Autism and Developmental Disorders, 37(8), doi: /s Schurz, M., Aichhorn, M., Martin, A., & Perner, J. (2013). Common brain areas engaged in false belief reasoning and visual perspective taking: a meta-analysis of functional brain imaging studies. Frontiers in Human Neuroscience, 7. doi: Artn 712 Doi /Fnhum Senju, A. (2012). Spontaneous Theory of Mind and Its Absence in Autism Spectrum Disorders. Neuroscientist, 18(2), doi: Doi / Surtees, A., Apperly, I., & Samson, D. (2013a). Similarities and differences in visual and spatial perspective-taking processes. Cognition, 129(2), Surtees, A., Apperly, I., & Samson, D. (2013b). The use of embodied self-rotation for visual and spatial perspective taking. Frontiers in Human Neuroscience, 7. doi: Artn 698 Doi /Fnhum Tan, J., & Harris, P. L. (1991). Autistic children understand seeing and wanting. Development and Psychopathology, 3(02), doi: doi: /s

27 26 Tessari, A., & Rumiati, R. I. (2004). The strategic control of multiple routes in imitation of actions. Journal of Experimental Psychology-Human Perception and Performance, 30(6), doi: Doi / Wellman, H. M., & Liu, D. (2004). Scaling of theory-of-mind tasks. Child Dev, 75(2), doi: DOI /j x Wimmer, H., & Perner, J. (1983). Beliefs about beliefs: representation and constraining function of wrong beliefs in young children's understanding of deception. Cognition, 13(1), doi: (83) [pii] Yirmiya, N., Sigman, M., & Zacks, D. (1994). Perceptual Perspective-Taking and Seriation Abilities in High-Functioning Children with Autism. Development and Psychopathology, 6(2), doi: Doi /S Yu, A., & Zacks, J. (2010). The role of animacy in spatial transformations. Memory & Cognition, 38(7), doi: /mc Zacks, J. M., & Tversky, B. (2005). Multiple Systems for Spatial Imagery: Transformations of Objects and Bodies. Spatial Cognition & Computation: An Interdisciplinary Journal, 5(4),

28 Tables and Table Legends Table 1: Participant demographics. All data are given as mean (±standard deviation) and range. N Age VMA BPVS Raw SCQ SAS ToM ASC 30 (27 male) 9.03±2.45 ( ) 6.55±2.19 ( ) 69.87±18.55 (46-119) 11.07±7.3 (0-30) 9.89±5.43 (2-27) 12±6.39 (2-33) TD 30 (18 male) 6.83±1.66 ( ) 6.68±2.12 ( ) 70.67±18.70 (40-120) ±4.45 (18-36) - t(58)=- 4.65,p<0.001 t(58)=0.23, p=0.82 t(58)=0.16, p=0.87 t(56)=10.99, p<0.001

29 28 Table 2: Results of the regression analyses for the TD and ASC groups. Beta values, confidence intervals, t values and p values are displayed for each variable. Bold values highlight significant predictors of VPT2 ability. Typical ASC (model 1) ASC (model 2) N N=30 N=30 N = 30 Overall model fit R²=0.65 R²=0.73 R²=0.78 MR performance ( ) ( ) ( ) t=0.593 p=0.558 Body Representation ( ) t=2.77 p=0.011 SAS ( ) t= p=0.683 BPVS raw score ( ) t=2.20 p=0.038 Age ( ) t=0.102 p=0.920 t=4.48 p= ( ) t= p= ( ) t= p= ( ) t=2.82 p= ( ) t= p=0.366 t=4.22 p= ( ) t= p= ( ) t=0.550 p= ( ) t=2.41 p= ( ) t= p=0.470 ToM Not included Not included ( ) t=0.848 p=0.406 SCQ Not included Not included ( ) t=-2.10 p=0.048

30 29 Figure Legends Figure 1: Examples of stimuli and tasks. A depicts the toy place on the turntable and an example of a response card given to the child. The toy is then covered. B depicts VPT2S: What will YOU see? C depicts VPT2O: What will JIM see? D depicts the mental rotation task, in which the toy is rotated and the child is asked which view they will see when the pot is lifted. Figure 2: An example of a trial in the body posture representation task, with exemplar, target and foil stimuli. A displays an example of a meaningful trial and B displays an example of a meaningless trial Figure 3: Mean scores (±standard error) for the TD and ASC children across the VPT and MR tasks. Each child completed 6 trials so the maximum score for each task was 6 and chance 1.5 (25%). Results are displayed here as a percentage. Figure 4: Mean scores (±standard error) for the TD and ASC children in the Meaningful and Meaningless body representation tasks. Each child completed 6 trials so the maximum score for each task was 6 and chance 1.5. Results are displayed here as a percentage. Figure 5: This figure displays the expected age appropriate performance across different domains in the typical and autistic children. Whilst the TD children are performing at the level appropriate for their chronological age across domains, the children with ASC are only performing at the expected level in the mental rotation task. Their performance on the VPT2 task is in line with their VMA, whilst performance on the body representation task is lower than expected for both chronological age and VMA. The arrows denote the results of the regression in both groups, showing a significant relationship between body representation and VPT2 in the TD group, and MR and VPT2 in the ASC group.